use_variable_labels

Validate/use variable labels from tabular file header

Specification

• Alias: None

• Arguments: None

Description

When importing global surrogate training data (or challenge evaluation points) from a tabular data file containing a header ( annotated or custom_annotated header), this keyword toggles validation of variable labels present in the header row.

The labels are validated against the descriptors of the variables being imported to. If the tabular file labels can be rearranged to match the expected labels, the columns in the data file will be reordered on read to match the Dakota variable order. If the read labels are not a permutation of, nor equal to, the expected labels, an error will result.

Default Behavior

When not specified, variable labels will be read, but not strictly enforced or reordered. A warning will be issued if the variable labels are not as expected, and guidance offered if they can be permuted to match expectations.

Expected Output

Console output will be generated for any warnings, as well as to indicate whether any variable reordering is taking place.

Usage Tips

The use of this keyword is recommended when importing header-annotated tabular data files where the variables are appropriately labeled. Tabular files do not always contain response labels, so no attempt is made to disambiguate variable from response labels. The variable labels must appear contiguously in the header after any leading column IDs such as ‘eval_id’ or ‘interface’.

Examples

This example enforces variable labels for both build and challenge points

model
  id_model = 'SURR'
  surrogate global
    polynomial quadratic
    import_build_points_file = 'dakota_surrogate_import.unc_fixedothers.dat'
      annotated  use_variable_labels
    challenge_points_file = 'dakota_surrogate_import.dat'
      annotated  use_variable_labels